Victoria Abril, Begoña Manuel-y-Keenoy, Rosa Solà, Jorge Luis García, Celeste Nessier, Rosendo Rojas, Silvana Donoso, and Victoria Arija.

Transcription

1 Prevalence of overweight and obesity among 6- to 9-year-old schoolchildren in Cuenca, Ecuador: Relationship with physical activity, poverty, and eating habits Victoria Abril, Begoña Manuel-y-Keenoy, Rosa Solà, Jorge Luis García, Celeste Nessier, Rosendo Rojas, Silvana Donoso, and Victoria Arija Abstract Background. The high prevalence of childhood overweight and obesity is a health problem worldwide. In developing countries, we lack information on the extent of the problem and the risk factors involved. Objective. To determine the prevalence rates of overweight and obesity and of abdominal obesity, and their relationship with physical activity, poverty, and eating habits in schoolchildren in Cuenca, Ecuador. Methods. A cross-sectional survey in a representative sample (n = 743) schoolchildren aged 6 to 9 years was conducted. Overweight and obesity were detected using the International Obesity Task Force cutoffs according to body mass index (BMI), and abdominal obesity was detected according to waist circumference. Poverty, physical activity, and eating habits were assessed with validated questionnaires. Results. The prevalence rates of overweight and obesity and of abdominal obesity were 26.0% and 10.6%, respectively. There were no differences between the sexes, but the prevalence of overweight and obesity was 1.5- to 2-fold higher in 9-year-old than in 6-year-old children (p <.05). Multivariate models demonstrated that higher BMI and waist circumference were significantly related to low physical activity and nonpoverty. Insufficient physical activity (in 75% of children) was associated with a 13% to 18% increased risk of overweight and obesity and abdominal obesity. Eating breakfast and eating Victoria Abril is affiliated with the University of Cuenca, Ecuador, and the Universitat Rovira i Virgili, Tarragona, Spain; Jorge Luis García, Rosendo Rojas, and Silvana Donoso are affiliated with the University of Cuenca; Victoria Arija and Rosa Solà are affiliated with the Universitat Rovira i Virgili; Begoña Manuel-y-Keenoy is affiliated with the University of Antwerp, Belgium; Celeste Nessier is affiliated with the Instituto de la Salud Juan Lazarte, Universidad Nacional de Rosario, Argentina. Please direct queries to the corresponding author: Victoria Arija, Nutrition and Public Health Unit, Universitat Rovira i Virgili, C/ Sant Llorenç 21, Reus, Spain; victoria.arija@urv.cat. more than three meals per day (in 96.7% and 85.9% of children, respectively) were not related to the prevalence of overweight and obesity. Eating fruits during school break was associated with a lower BMI. Conclusions. The high prevalence of overweight and obesity observed in schoolchildren increased from the ages of 6 to 9 years and was associated with insufficient physical activity and nonpoverty. Promoting physical activity and fruit consumption in school snacks should be explored as intervention measures to prevent and reduce overweight and obesity in Cuenca schoolchildren. Key words: Eating habits, nutritional status, obesity, overweight, poverty, schoolchildren Introduction A high prevalence of overweight and obesity is observed in developing and developed countries [1, 2]. The prevalence of overweight and obesity in the United States reported in 2009/10 was 26.7%, 32.6%, and 33.6% in children 2 to 5, 6 to 11, and 12 to 19 years of age, respectively [2]. The prevalence of overweight and obesity was 11%, 39.3%, 35.5%, and 41.8% in 5- to 12-year-old Colombian children, 9- to 12- yearold Chilean children, 10- to 11-year-old Argentinian children, and 5- to 14-year-old Mexican children, respectively [3 6]. There are few data on childhood prevalence of overweight and obesity. In 2008, a study in 12- to 18-year-old adolescents in Ecuador showed a prevalence of 13.7% for overweight and 7.5% for obesity [7]; more recently, a representative survey of 10- to 16-year-old adolescents in Cuenca, Ecuador, revealed a prevalence of 18% for overweight and 2.1% for obesity [8]. Childhood obesity increases the risk of adult obesity as well as chronic cardiometabolic diseases such as type 2 diabetes, hypertension, and cardiovascular disease that present an important public health burden in many countries [1, 9]. More specifically, detecting 388 Food and Nutrition Bulletin, vol. 34, no , The Nevin Scrimshaw International Nutrition Foundation.

2 Overweight and obesity in Ecuadorian children abdominal obesity by measuring waist circumference is considered useful to assess cardiometabolic risk [10, 11]. Abdominal obesity has been shown to be related to hyperinsulinemia, insulin resistance, hypertension, and impaired fasting glucose [12 15]. There has been an important increase of noncommunicable chronic diseases in Ecuador in the past decades. The number of people with type 2 diabetes mellitus and hypertension increased about 10 times from 1994 to 2010 [9]. Some dietary practices and sedentary lifestyles predispose to increased weight. Overweight and obesity are associated with a lower number of meals per day [16], low consumption of fruits and vegetables [17], skipping breakfast, energy expenditure that is lower than energy intake, screen time (television viewing and computer use), and other sedentary behavior [16, 18 25]. Children who play sports three or more hours per week have a lower risk of developing noncommunicable diseases and have less abdominal fat than children who only play sports 90 minutes or less per week [26]. Low socioeconomic status is also associated with the high prevalence of overweight and obesity in children in developing and developed countries [6, 19]. Information about the nutritional status of schoolchildren in Ecuador is incomplete, and the risk factors that favor the development of overweight and obesity in this context need to be identified. The aim of this study was therefore to determine the prevalence of overweight and obesity in 6- to 9-year-old schoolchildren in the city of Cuenca, Ecuador, and to analyze the relationship of overweight and obesity with potential risk factors such as physical activity, poverty, and eating habits. Methods Study population In this cross-sectional study, data were collected from September 2010 to March 2011 from 6- to 9-year-old schoolchildren in Cuenca, Ecuador. The representative sample was drawn from a population of 35,974 children with the use of the statistical software EpiInfo, version Accepting a confidence of 95% and a beta risk of 0.2 with a bilateral contrast, and assuming a prevalence of 25% of overweight and obesity that was reported in studies on adolescents in Ecuador [7], a sample size of 624 children was calculated. Assuming 25% lack of participation, the final sample needed was estimated to be 832 children. The schools of the urban area of Cuenca were selected for conglomerates taking into account the type of school (public, private. or subsidized) and 15 public schools, 13 private schools, and 2 subsidized private schools were selected randomly. The participants were chosen randomly from the lists of children attending 389 the second to the fifth level that were provided by the authorities of each school. Random selection of schools by conglomerates and of the children within each level in each school ensured that our study sample was representative of the population group of schoolchildren 6 to 9 years of age, because attendance at primary school is mandatory in this country. Besides, a high level of participation (89.3%) was achieved with repeated visits to schools to complete data collection from all children whose parents gave informed consent. It should be noted that although the selection process was as described, schools in Cuenca are imbalanced, with more of one sex than the other; the ratio of male to female students in the schools selected turned out to be 0.65, which is not representative of the school population of Cuenca as a whole. Children with physical deformities in the hips or legs that prevented them from standing or from undergoing anthropometric measurement were excluded. This exception applied to one boy who had a deformity in the hip that prevented him from standing and who was not included in our study. We returned to the schools whenever it was necessary to complete the measurements in all children whose parents had signed the informed consent form. The Bioethics Committee of Cuenca University approved the study protocol (002-C0Bl-10). Informed consent was signed by the parents. The trial was conducted according to the Declaration of Helsinki and good clinical practice guidelines of the International Conference of Harmonization. The study was registered on the website of Clinical Trials (NCT ). Anthropometric measurements Height and weight were measured in duplicate by two trained interviewers using a standard protocol. Body weight was measured to the nearest 0.1 kg by a calibrated SECA digital scale with the children wearing light clothing. Height was measured to the nearest 1 mm by a Health O Meter PORTROD portable stadiometer with the children not wearing shoes. The World Health Organization (WHO) 2007 growth reference curves were used to define nutritional status. With the use of WHO Anthro 2007 software, weight and height data were used to calculate z-scores, which indicate how far (in SD units) a measurement is from the mean [27] weight-for-age, height-for-age, and body mass index (BMI)-for-age [28]. Nutritional status was defined as follows: underweight (weight-for-age z-score below the 2 z-score line), wasted (BMI-forage z-score below the 2 z-score line), normal weight (weight-for-age or BMI-for-age z-score between the 2 and +1 z-score lines), overweight (weight-for-age or BMI-for-age z-score between the +1 and +2 z-score lines), obesity (weight-for-age or BMI-for-age z-score above the +2 z-score line), stunted (height-for-age z-score below the 2 z-score line), and normal height

3 390 V. Abril et al. (height-for-age z-score between the 2 and +2 z-score lines). As an alternative reference to assess nutritional status, we also used the International Obesity Task Force (IOTF) BMI cutoffs available in Cole s tables [29, 30]. Percentile curves that correspond to the BMI cutoff points at the age of 18 years were used to define underweight (below 18.5 kg/m 2 ), normal weight (between 18.5 and 24.9 kg/m 2 ), overweight (between 25.0 and 29.9 kg/m 2 ), and obesity (30.0 kg/m 2 or above). Waist circumference was measured midway between the lowest rib margin and the iliac crest with the subject in a standing position and was recorded to the nearest 1 mm. A nonelastic flexible SECA measuring tape for circumferences was employed. Percentiles of waist circumference were based on the National Health and Nutrition Examination Survey (NHANES III) data, stratified for age and sex. Abdominal obesity was defined as having a waist circumference at or above the 90th percentile for age and sex [31]. Physical activity habits The quality of physical activity habits was assessed with a specially designed questionnaire validated in Chile in children from 6 to 10 years old [26]. The responses to the questionnaires were qualified using a numeric scale that ranged from 0 to 10 points. A higher score indicated a better habit (more physical activity). The test assayed the number of hours spent lying per day, the number of sedentary hours per day, the number of house blocks walked per day, the number of hours of sports per day, and the number of hours of sports per week [26]. The total scores were classified according to tertiles: first tertile, 0.0 to 3.0; second tertile, 3.01 to 4.00; and third tertile, 4.01 to Poverty The questionnaire used to define poverty was developed by the Integrated System of Social Indicators of Ecuador of 2003 (SIISE 2003). It defines a household as poor if it has one or more of the following deprivations: 1, house without electricity; 2, no running water the house is supplied with river water, rain water, or nondrinking water; 3, no toilet or latrine; 4, house with four people or more per bedroom (not including kitchen, bathroom, and garage); 5, parent with 3 or fewer years of schooling; 6, one or more household members aged 10 years or more are illiterate; 7, one or more members between 7 to 12 years of age are not enrolled in an educational institution; 8, the recent birth of one of its members did not receive medical care; 9, one or more of its members were unable to perform their normal activities for 3 days or more due to illness or accident and were not seen by a doctor, nurse, or pharmacist; 10, the ratio of the total number of members to the number of members working is greater than or equal to 4 [32]. We did not include question number 7 due to lack of information in the questionnaires. Eating habits Three direct questions were asked to obtain information about habits such as skipping breakfast, number of meals a day, and eating fruit during the break at school. The number of meals included breakfast, lunch, dinner, and two snacks during the day (one in the middle of the morning and one in the middle of the afternoon according to whether the child goes to school in the morning or the afternoon). In addition, children can take snacks at home when they are not at school. This adds up to a total of five meals per day. Appendix 1 shows the questionnaire on eating habits that we used. Statistical analysis Data were entered in duplicate in EPIDATA 3.1 by two independent researchers and cross-checked for errors. Analyses were performed with SPSS 18 for Windows software. Descriptive statistics were given as frequencies and 95% confidence intervals, means, standard deviations, medians, and interquartile ranges according to normality of variables (Shapiro Wilks p >.05). The chi-square test was applied to analyze differences in the frequency distribution of categorical variables. To test between-group differences in quantitative variables, Student s t-test and analysis of variance (ANOVA) were performed. Correlation was analyzed with the Spearman rank correlation test. A p value <.05 was considered to indicate a statistically significant difference. Multivariate linear regression models were used to analyze the relation between BMI and waist circumference measurement as dependent variables, with sex (female = 0, male = 1), age (years), physical activity score (0 to 10 points), father s educational level (years of study), mother s educational level (years of study), poverty (nonpoor = 0, poor = 1), breakfast (every day = 0, sometimes or never = 1), number of meals per day (four or five = 0, three or fewer = 1), and eating fruit at school break (yes = 0, no = 1) as independent variables. Logistic regression models were used to identify the variables that predicted the presence or absence of overweight and obesity and abdominal obesity (no = 0, yes = 1) (dependent variables) by analyzing the same independent variables as in the linear regression models. Results Characteristics of the study population Of the total sample, 743 children (89.3%) participated in our study. The descriptive statistics of children aged

4 Overweight and obesity in Ecuadorian children 6 to 9 years (mean ± SD, 7.9 ± 1.1 years) according to sex are summarized in table 1. The proportion of the sample for each age group was similar: 25.2%, 25.7%, 25.8%, and 23.3% for 6, 7, 8, and 9 years, respectively. However, there was a higher proportion of girls (60%) than of boys in our sample. We excluded one boy who 391 had a deformity in the hip that prevented him from standing. We found a positive correlation (r = and p <.001) between BMI and waist circumference. There were differences in weight and height between all age groups. BMI and waist circumference did not differ between children 7 and 8 years of age (p >.05). TABLE 1. Anthropometric characteristics of schoolchildren from 6 to 9 years old by sex and age a Characteristic Sex Age group (yr) Male Female p b n % Age (yr) 7.9 ± ± ± ± ± ± 0.2 < ± 1.1 Weight (kg) 26.1 ± ± ± ± ± ± 6.6 < ± 6.2 Height (cm) ± ± ± ± ± ± 5.7 < ± 8.2 BMI (kg/m 2 ) 17.1 ± ± ± ± ± ± 2.9 < ± 2.5 WC (cm) 60.5 ± ± ± ± ± ± 9.2 < ± 3 BMI, body mass index (weight/height 2 ); WC, waist circumference a. Plus minus values are means ± SD. b. Student s t-test. c. ANOVA test. p c Total TABLE 2. Nutritional status of schoolchildren in Cuenca, Ecuador (n = 743) Indicator Sex Age group (yr) Male Female p 6 (a) 7 (b) 8 (c) 9 (d) p *p between groups Total % (95% CI) Height-for-age a Stunting (%) ( ) Normal height (%) ( ) Weight-for-age a Normal weight (%) ( ) Overweight and obesity (%) *p (a b) =.004 *p (a d) < ( ) BMI b Normal weight (%) ( ) Overweight and *p (a d) =.019 obesity (%) *p (c d) 32.7 ( ) =.048 BMI c Normal weight (%) ( ) Overweight and *p (a d) = ( ) obesity (%) Abdominal obesity d No (%) ( ) Yes (%) *p (a d) = ( ) BMI, body mass index (weight/height 2 ) a. In the first analysis, we found a significant p-trend value inside some groups (weight for-age, BMI, and waist circumference). Then we analyzed in pairs the differences between the different age groups (a with b, a with c, and a with d, and so on for the different ages. We found significant differences in anthropometric measurements between a and b, a and d, and c and d. The chi-square test was used to calculate p and *p between groups. b. According to WHO criteria (2007). c. According to IOTF criteria (2000). d. According to NHANES III criteria.

5 392 V. Abril et al. Nutritional status by sex and age As shown in table 2, the prevalence of stunting was 6.4%. We found 2.2% of children to be underweight according to the WHO criteria for weight-for-age cutoff, and 0.1% and 3.6% of subjects to be wasted children according to the WHO and IOTF cutoffs, respectively. Overweight and obesity according to weight-for-age was observed in 20.1% of children. The prevalence of overweight and obesity was higher in the 9- and 7-year-old children, 26.6% and 23.2% respectively, than in the 6-year-olds, 12.0% (p =.003). The prevalence of overweight and obesity according to BMI-for-age was 32.7% (WHO criteria) and 26% (IOTF criteria) in the total group. Again, it was higher in the 9-year-olds (40.1% by the WHO criteria and 32.0% by the IOTF criteria) than in the 7-year-olds (33.0% by the WHO criteria and 25.0% by the IOTF criteria) (p <.05). There were no statistically significant differences between boys and girls. Abdominal obesity was observed in 9.5% of boys and 11.4% of girls. It tended to be higher in the 9 year-olds (15.7%) and in the 7 year-olds (10.5%) (p =.079). The proportion of children with overweight and obesity increased with increasing age (p-value for trend.003,.042, and.028 for weight-for-age, BMI-for age according to WHO, and BMI-for age according to IOTF, respectively). Anthropometric measurements were significantly different between 6- and 7-year-olds, 6- and 9-year-olds, and 8- and 9-year-olds (table 2). Physical activity habits Results on physical activity are summarized in table 3. The median of the total score for physical activity was 4 (percentile 25 = score 3, percentile 75 = score 5). Insufficient physical activity (defined as having a score in the lowest tertile (0 to 3 points) was observed in 45.4% TABLE 3. Physical activity of schoolchildren according to sex, age, overweight and obesity, and poverty a Variable 1st tertile ( points) 2nd tertile ( points) 3rd tertile ( points) Mean (SD) 2.5 (0.6) 4.0 (0.0) 5.6 (0.7) % schoolchildren (95% CI) ( ) ( ) ( ) Sex Male (%) Female (%) Age (yr) 6 (%) (%) (%) (%) BMI c Normal weight (%) Overweight and obesity (%) BMI d Normal weight (%) Overweight and obesity (%) Abdominal obesity e No (%) Yes (%) Poverty f No (%) Yes (%) BMI, body mass index (weight/height 2 ) a. The physical activity test was validated with a numeric scale from a minimum of 0 (less physical activity) to a maximum of 10 (more physical activity) points. b. The chi-square test was used to calculate p between groups. c. According to WHO criteria (2007). d. According to IOTF criteria (2000). e. According to NHANES III criteria. f. According to SIISE criteria (2003). p b

6 Overweight and obesity in Ecuadorian children of children and was not related to sex, age, or prevalence of overweight and obesity, abdominal obesity, or poverty. Overall, 75% of these children played sports 4 or fewer hours per week (percentile 75 = 4 hours) and spent 8 hours or more daily sitting (time at school, television viewing, Internet, and computer games). Poverty As shown in table 4, 26.2% of this group of children were classified as living in poverty; living in poverty was not related to sex or age group. Overweight and obesity tended to be more prevalent in nonpoor children (according to the two BMI criteria, p =.062 and.078), but this difference did not reach statistical significance when the groups with and without abdominal obesity were compared. Eating habits Table 5 shows that 96.7% of children ate breakfast daily and that 85.9% ate four or five meals per day. These proportions did not differ significantly between boys and girls, children in different age groups, or children with and without overweight and obesity. Some 59.1% of children ate fruit during the school break, and the proportion was significantly higher in girls (62.3%) than in boys (54.4%, p =.034). Analysis of lifestyle risk factors In order to detect the lifestyle factors predictive of nutritional status and the presence of overweight and obesity in this group of children, we conducted multivariate regression analysis using two approaches. First, we analyzed two linear regression models, one with BMI and one with waist circumference as the dependent variable (table 6). Then we analyzed three logistic regression models with the presence of overweight and obesity classified according to two cutoff criteria (WHO and IOTF) and the presence of abdominal obesity as dependent variables (table 7). In the linear regression models (table 6), the factors studied explained 8.2% and 18.1% of the variability of BMI and waist circumference, respectively. Age was positively related to both BMI and waist circumference (p <.001). Significant negative relationships were observed for both physical activity (p =.009 and p =.016 for BMI and waist circumference, respectively) and poverty (p =.012 and p =.040, respectively). Eating fruit during break at school was significantly associated with a lower BMI (p =.027) and tended to be associated with a lower waist circumference (p =.072). Using logistic regression models analyzing the same independent variables as in the linear regression models, we detected the factors that might be predictive of the presence of overweight and obesity (according to 393 TABLE 4. Poverty in schoolchildren according to sex, age, and overweight and obesity Variable % schoolchildren (95% CI) the WHO or the IOTF criteria) and abdominal obesity (table 7). As indicated by the odds ratios, children were 13% to 15% (p <.05) less likely to be overweight and obese and 18% (p =.048) less likely to have abdominal obesity if they had a high physical activity score. Poor children were 35% less likely to be overweight and obese according to the WHO criteria (p =.05) and 39% less likely to be overweight and obese according to the IOTF criteria (p <.05). In contrast, poverty did not predict abdominal obesity. Eating fruit during school break tended to be associated with a lower prevalence of overweight and obesity (p =.076) and abdominal obesity (p =.057). Discussion Poverty a Yes (%) No (%) 26.2 ( ) 73.8 ( ) Sex Male (%) Female (%) Age (yr) 6 (%) (%) (%) (%) BMI c Normal weight (%) Overweight and obesity (%) BMI d Normal weight (%) Overweight and obesity (%) Abdominal obesity e No (%) Yes (%) BMI, body mass index (weight/height 2 ) a. Poverty was determined according to the SIISE (2003) criteria. b. The chi-square test was used to calculate p between groups. c. According to WHO criteria (2007). d. According to IOTF criteria (2000). e. According to NHANES III criteria. This study assessed nutritional status in a sample of schoolchildren representative of the urban area of Cuenca, Ecuador, in response to the need for data from WHO on childhood nutritional status in this country. It also investigated the relationship of overweight p b

7 394 V. Abril et al. TABLE 5. Eating habits of schoolchildren according to sex, age, and overweight and obesity Variable % schoolchildren (95% CI) Every day 96.7 ( ) Breakfast (%) No. of meals per day (%) Eats fruit at school break (%) Sometimes or never p a 4 or 5 3 p a Yes No p a 3.3 ( ) 85.9 ( ) 14.1 ( ) 59.1 ( ) 40.9 ( ) Sex Male (%) Female (%) Age (yr) 6 (%) (%) (%) (%) BMI b Normal weight (%) Overweight and obesity (%) BMI c Normal weight (%) Overweight and obesity (%) Abdominal obesity d No (%) Yes (%) BMI, body mass index (weight/height 2 ) a. The chi-square test was used to calculate p between groups. b. According to WHO criteria (2007). c. According to IOTF criteria (2000). d. According to NHANES III criteria. and obesity with physical activity, poverty, and eating habits. Our study established that the prevalence of overweight and obesity is high in this group and is related to the presence of risk factors, such as insufficient physical activity, nonpoverty, and not eating fruit snacks in school. The results show that the prevalence varies depending on the cutoff criteria used to define overweight and obesity, ranging from 32.7% (WHO BMI criteria) to 26.0% (IOTF BMI criteria) to 20.1% (WHO weightfor-age criteria). We consider that the WHO cutoffs are better for our population group because they are applicable to children of the entire world independently of ethnicity or socioeconomic status and with favorable conditions for growth of children and control of infections [33]. Moreover, WHO uses three criteria: weight-for-age, height-for-age, and BMI-for-age. This allows a better and more complete assessment of nutritional status at the moment and can give information about chronic malnutrition in cases of stunting. Despite the difficulties of comparing surveys from different countries (with different age groups and various cutoff criteria to define overweight and obesity), it is clear that the prevalence of overweight and obesity in this age group of Cuencan schoolchildren is in the same range as those in the aforementioned studies in schoolchildren in some Latin American countries [4, 5]. Nonetheless, the prevalence of overweight and obesity found is lower than that in Mexico (41.8%) [6], higher than that in Bogotá, Colombia (11%) [3], and similar to that in the United States (32.6%) [2]. We did not observe any significant differences between girls and boys, contrary to other studies that found higher prevalences of obesity in males than in females in the same group [2, 4]. This difference may be due to the use of other criteria (Centers for Disease Control and Prevention) to define nutritional status. It is also possible that in the other studies the cutoff values used for girls were not discriminant enough to classify them as overweight or obese. As a consequence, fewer girls

8 Overweight and obesity in Ecuadorian children 395 TABLE 6. Relationship of lifestyle risk factors with body mass index and waist circumference in schoolchildren Risk factor BMI Multiple linear regression Waist circumference (cm) β SE p β SE p Sex (female = 0, male = 1) Age (yr) < <.001 Physical activity test (score) a Poverty (nonpoor = 0, poor = 1) b Father s education (yr) Mother s education (yr) Daily breakfast (yes = 0, no = 1) Meals per day (4 or 5 = 0, 3 = 1) Eats fruit at school break (yes = 0, no = 1) (Constant) < <.001 F = 6.91; p <.001; R 2 C.100 = 8.2% F = 15.54; p <.001; R 2 C.100 = 18.1% BMI, body mass index (weight/height 2 ) a. The physical activity test was validated with a numeric scale from a minimum of 0 (less physical activity) to a maximum of 10 (more physical activity) points. b. Poverty was determined according to the SIISE (2003) criteria. TABLE 7. Relationship of lifestyle risk factors with the presence of overweight and obesity and abdominal obesity in schoolchildren Risk factor Logistic regression Overweight and obesity a Overweight and obesity b Abdominal obesity c OR (95% CI) p OR (95% CI) p OR (95% CI) p Sex (female = 0, male = 1) 1.19 ( ) ( ) ( ).126 Age (yr) 1.16 ( ) ( ) ( ).095 Physical activity test (score) 0.87 ( ) ( ) ( ).048 Poverty (nonpoor = 0, poor = 1) d 0.65 ( ) ( ) ( ).417 Father s education (yr) 1.00 ( ) ( ) ( ).903 Mother s education (yr) 0.99 ( ) ( ) ( ).780 Daily breakfast (yes = 0, no = 1) 1.05 ( ) ( ) ( ).920 Meals per day (4 or 5 = 0, 3 = 1) 1.06 ( ) ( ) ( ).558 Eats fruit at school break (yes = 0, no = 1) 1.17 ( ) ( ) ( ).057 a. According to WHO criteria (2007). b. According to IOTF criteria (2000). c. According to NHANES III criteria. d. According to SIISE criteria (2003). than boys were defined as overweight or obese [4]. Nevertheless, our observation is in the same line as the study in adolescents in the same urban area that did not find differences in categories of BMI by sex [8] and the study in Bogotá, Colombia [3], that used the same criteria as in our study. An interesting observation is that the prevalence of overweight and obesity increased with age and was 1.5 to 2 times higher at the age of 9 than at 6 years. This age-dependent increase in childhood overweight and obesity is also observed in other studies that describe an increase of BMI in children around 6 years of age [34 36] and a higher prevalence of obesity at 6 to 9 years than at 5 years or less [3]. A study in India showed a rapid increase in BMI in children through 12 years of age [37]. Another study observed that the prevalence of overweight and obesity increased from 8 to 12 or 13 years and decreased thereafter [6]. This seemed to be explained by the increased self-consciousness among postpubertal children about weight gain and physical appearance [6]. In contrast, in the United States, the high prevalence of overweight and obesity in childhood persists during the adolescent years [2]. The parallel study on 10- to 16-year-olds in Cuenca reported a prevalence of 21.3% of overweight and obesity among the urban adolescents according to the IOTF cutoffs

9 396 V. Abril et al. [8], which is lower than the prevalence observed in schoolchildren in our study. This early increase of BMI in our group of young children could be a consequence of a rapid nutritional transition that is currently affecting the younger children (and their mothers) more intensely than some years ago when the cohort of current adolescents were young children [38]. This process was also observed among urban children in Chile and in Ecuador who commonly eat Western-style food and have a higher consumption of sugar-sweetened beverages and energy-dense fast foods, potato chips, etc. [17, 38]. This positive energy imbalance is worsened by the limited availability of open spaces and parks [6]. The prevalence of abdominal obesity was 10.6% in this group of children. It increased with age, reaching 15.6% among 9-year-olds. In 10- to 16-year-old adolescents in the same city, the prevalence was 19.5% [8]. It should be noted, however, that the cutoff criteria to define abdominal obesity differed (80th percentile in the adolescent study and 90th percentile in our study). Nevertheless, these prevalence figures were lower than those observed in urban adolescents in Mexico (27.9%) [39] and Brazil (32.7%) and higher than those in Argentina (13.9%) [39]. Obesity prevention campaigns to change lifestyle habits through changes in environment such as the school are found to be more successful, especially if they take place early in life [40, 41]. The gradual rise in BMI in children aged 6 to 9 years revealed by our results clearly indicates the need to start prevention early during primary school. The reasons for this increase are not clear and need to be corroborated by following up the same cohort from 6 to 9 years of age. It could be due to an adiposity rebound that corresponds to a rise in the BMI curve that occurs between 5 and 7 years of age [38]. It is thus especially relevant to identify the factors determining overweight and obesity in this age group and in the specific context of an urban setting in a country in transition such as Ecuador. First of all, we observed that these children had sedentary habits that are similar to those among children in other countries [19, 26]. This situation has been explained by the greater access to computers and television and the increasing use of school buses or cars compared with some years ago [42]. We observed that 75% of the children played sports for 4 or fewer hours per week. This is considerably less than the 60 minutes or more of moderate- to vigorous-intensity physical activity daily proposed in Global Recommendations of Physical Activity for Health by WHO for this age group to provide health benefits [43]. Indeed, multivariate regression analyses revealed that high physical activity was a significant independent factor determining a lower BMI and waist circumference and decreasing the risk of overweight and obesity and abdominal obesity by 13% to 18%. In the univariate analysis (chi-square tests) shown in table 3, we did not find any significant differences according to sex, age group, poverty group, or adiposity group when comparing the children in the three tertiles of physical activity. This artificial grouping using rigid cutoffs for these parameters and the skewed distribution (with 45% in the lowest tertile of physical activity) may explain the failure to detect any statistically significant differences between these groups. However, we did not use cutoffs for physical activity in the multivariate regression analysis, because we introduced the total score of the physical activity test as a continuous variable (0 to 10 points) in the regression model. Likewise, in the linear regression model, the parameters of adiposity (BMI and waist circumference) were analyzed as continuous variables, thus avoiding the use of cutoffs and possible misclassification into the overweight and obese or the normal weight groups. Our findings are agreement with those of another study conducted in Rome, Italy, which found an association between lower physical activity and overweight in children [16]. A study found that moderately active boys were twice as likely to be of normal weight (OR = 2.10; 95% CI, 1.07 to 3.99; p <.003) [44]. Physical activity was associated with significantly lower values of waist circumference (100 counts per minute of total physical activity, determined by accelerometer, decreases waist circumference by 0.35 cm) [45]. Furthermore, physical activity in children and adolescents decreases systolic blood pressure, decreases serum levels of insulin and triglycerides, and increases HDLcholesterol [45]. This observation again underscores the importance of providing and stimulating physical activity early on in school life. Interestingly, our data showed that poor children had a lower BMI than nonpoor children. Poor children were 35% to 39% less likely to be overweight and obese than nonpoor children. However, nonpoverty could not predict abdominal obesity significantly. The reasons for this surprising divergence will need to be unraveled by more detailed studies in this study group. The relationship between socioeconomic status and the prevalence of overweight and obesity has also shown divergent results in other studies [19, 46]. Whereas in developed countries childhood obesity is inversely related to socioeconomic level, in developing countries there is a higher prevalence of overweight in children studying in private schools or with better economic status [6, 7], as seen in our study. Studies in other countries determine socioeconomic status by type of school (public or private) [6, 19], ownership of a car, ownership of computers, own bedroom for the child or holidays abroad [21], and parents level of education and income [47 49]. In contrast, in our study we included questions about having electricity or running water at home that are no longer appropriate for defining low socioeconomic status in developed countries where these facilities are available at all levels of society. These considerations are

10 Overweight and obesity in Ecuadorian children relevant when comparing cross-country studies on the impact of poverty. Skipping breakfast was a significant independent predictor of being overweight in schoolchildren in Pakistan (OR = 1.82; 95% CI, 1.22 to 2.71) [50], and the BMI of children in New Zealand who skipped breakfast was 1.28 kg/m 2 higher than in those who usually had breakfast [49]. Adolescents in Taiwan with irregular consumption of breakfast had a 51% higher risk of being overweight [51]. A lower number of meals consumed per day is also associated with a higher prevalence of overweight (p <.001), although this association is still controversial [16]. We did not observe any significant relationships with these eating habits, but it should be noted that in our sample most children ate breakfast. Nevertheless, we observed a lower BMI in children who ate fruit during the school break, but only a tendency toward lower waist circumference and abdominal obesity. This observation is partly in agreement with other studies that have shown a negative association between consumption of fruits and vegetables and overweight [52]. One should note here that our study does not provide an answer to the question whether low fruit consumption was associated with a higher consumption of high-calorie foods such as sweet beverages and cakes, which might have a stronger impact on the deposition of abdominal fat. Other studies have associated low intake of fruit with high consumption of calorie-dense foods in all age groups [53]. Our linear regression models indicated that lifestyle factors, such as physical activity, poverty, and fruit eating, together with age, explained only 8.2% and 18.1% of the variability of BMI and waist circumference, respectively. This suggests that other factors that we did not analyze in our study, such as maternal prepregnancy overweight, high birthweight, or absence of breastfeeding, may have a stronger impact on the presence of overweight and obesity in schoolchildren [54]. For instance, the prevalence of obesity in children of obese mothers was higher than in children of mothers with BMI in the normal range [55]. Children with elevated birthweight ( 4.0 kg), had a higher risk of overweight at 6 to 13 years of age than children with a birthweight between 3.0 and 3.5 kg [56]. It is also important to note that the search for the lifestyle factors determining overweight and obesity in schoolchildren has several limitations in this crosssectional study design. It is recognized that comparisons of groups at different ages are not a substitute for a longitudinal cohort design in assessing age-dependent trends. Conclusions for true individual change cannot be derived from a transverse analysis of a population. Moreover, the higher proportion of girls than boys in our study population limited the power to detect any real differences associated with sex. In addition, the group studied was limited to schoolgoing children in 397 the urban setting of Cuenca, and our findings may not be extrapolated to rural populations and to populations in other areas of Ecuador. Some studies have shown variations in the prevalence of overweight and obesity among adolescents in different geographic regions in Ecuador (24.7% in the coast vs. 17.7% in the Sierra) [7]. There are differences between urban and rural areas of other countries [57], and in adolescents in the province of Azuay, the prevalence of overweight and obesity was higher in the urban area than in the rural area (21.3% vs. 15.7%) [8]. Furthermore, although it is estimated that 97% of children were registered in schools in Ecuador in , it is possible that some children do not go to school due to poverty [58]. This implies that a survey limited to schoolgoing children in the urban setting may exclude the most vulnerable group, those not able to go to school for socioeconomic reasons [58]. Another important limitation is the use of anthropometric cutoffs derived from data in other countries to define overweight and obesity and abdominal obesity in the Cuencan schoolchildren. The divergences in prevalence when different criteria are used suggest that they could overestimate or underestimate the prevalence of overweight and obesity and of abdominal obesity in these children [15, 59] and may not be adequate or sensitive to assess obesity risk factors in this specific group. These considerations underscore the need to investigate adequate cutoffs for the nutritional assessment of overweight and obesity, abdominal obesity, and the influence of poverty and other risk factors on the nutritional status of this population of children in Ecuador. In any case, the high prevalence of overweight and obesity points to the need for public health campaigns aimed at preventing and reducing overweight in the children of Ecuador. Safe and appropriate outdoor places are needed for children and their families where they can practice sports or other physical activities. Furthermore, eating and physical activity habits and inclinations to food are acquired in early childhood [60], and government educational policy should focus on these aspects. It is important to know that Ecuadorian girls and boys show overweight and obesity patterns similar to their parents, and therefore promotion of healthy lifestyles by mass media campaigns should address the whole family and not only the children [38]. Ecuador is a country with four geographic regions (Andes Mountains, Amazonian rainforest, Pacific coast, and Galapagos Islands) that are different from one another in landscape, temperature, culture, etc., and it is complicated to extrapolate the findings from one city to the whole country. For this reason it will be necessary to assess nutritional status in different age groups and in different cities in Ecuador before designing and implementing interventional programs to promote healthy lifestyles in the various population groups (38). In conclusion, we observed a high prevalence of

11 398 V. Abril et al. overweight and obesity, about 26%, in schoolchildren in Cuenca. The prevalence was 1.5 to 2 times higher in 9-year-olds than in 6-year-olds. A higher BMI was associated with insufficient physical activity and nonpoverty. Although most children ate breakfast and a sufficient number of meals per day, eating fruit during the school break was favorably associated with a lower BMI. However, it is necessary to design cohort studies to elucidate the factors that are associated with an increase in weight in this age group. In addition, the efficacy of campaigns to prevent obesity by promoting physical activity and fruit-based school snacks needs to be tested in intervention studies on young schoolchildren in the city of Cuenca. Authors contributions Victoria Abril is a Ph.D. student in the Universitat Rovira i Virgili. Victoria Arija and Rosa Solà are her Spanish doctoral thesis supervisors. Victoria Abril designed and conducted the research (project, conception, development of overall research plan, collected data, and study oversight). Jorge Luis García, Celeste Nessier, Rosendo Rojas, and Silvana Donoso collaborated in the project, its conception, development of the overall research plan, and study supervision in Ecuador. Victoria Abril, Victoria Arija, Begoña Manuel-y-Keenoy, and Rosa Solà analyzed the data or performed statistical analysis in Reus. Victoria Abril, Victoria Arija, Begoña Manuel-y-Keenoy, and Rosa Solà wrote the manuscript (authors who made a major contribution). The final manuscript was read and approved by all coauthors. Victoria Arija, Begoña Manuel-y-Keenoy, and Rosa Solà have responsibility for the contents of the manuscript. Acknowledgments The authors wish to acknowledge the financial support of the Directorate of Research of the University of Cuenca, Ecuador. We are grateful to the Secretaría Nacional de Educación Superior, Ciencia, Tecnología e Innovación (SENESCYT), Ecuador. The authors want to acknowledge the help of Marlene Gía, Delfa Capelo, and Ana Mejia. The funders had no role in the study design, data collection and analysis, decision to publish, or preparation of the manuscript. The authors declare that they have no competing interests. References 1. World Health Organization. Diet, nutrition and the prevention of chronic diseases. World Health Organ Tech Rep Ser 2003;916: Ogden CL, Carroll MD, Kit BK, Flegal KM. Prevalence of obesity and trends in body mass index among US children and adolescents, JAMA 2012; 307: McDonald CM, Baylin A, Arsenault JE, Mora-Plazas M, Villamor E. Overweight is more prevalent than stunting and is associated with socioeconomic status, maternal obesity, and a snacking dietary pattern in school children from Bogota, Colombia. J Nutr 2009;139: Liberona Y, Engler V, Castillo O, Villarroel L, Rozowski J. Ingesta de macronutrientes y prevalencia de malnutrición por exceso en escolares de 5 Y 6 básico de distinto nivel socioeconómico de la región metropolitana. Rev Chil Nutr 2008;35: Kovalskys I, Rausch Herscovici C, De Gregorio MJ. Nutritional status of school-aged children of Buenos Aires, Argentina: data using three references. J Public Health 2011;33: Gupta N, Goel K, Shah P, Misra A. Childhood obesity in developing countries: epidemiology, determinants, and prevention. Endocr Rev 2012;33: Yepez R, Carrasco F, Baldeón M. Prevalencia de sobrepeso y obesidad en estudiantes adolescentes ecuatorianos del área urbana. Arch Latinoam Nutr 2008;58: Ochoa-Avilez A, Andrade S, Huynh T, Verstraeten R, Lachat C, Rojas R, Donoso S, Manuel-y-Keenoy B, Kolsteren P. Prevalence and socioeconomic differences of risk factors of cardiovascular disease in Ecuadorian adolescents. Pediatr Obes 2012;7: Ministerio de Salud Pública del Ecuador. Salud del adulto. Enfermedades crónicas no transmisibles. Available at: content/article/49-programas/enfermedades-cronicasno-transmisibles/157-salud-del-adulto-nfermedadescronicas-no-transmisibles. Accessed 9 September Lee S, Bacha F, Gungor N, Arslanian SA. Waist circumference is an independent predictor of insulin resistance in black and white youths. J Pediatr 2006;148: Hitze B, Bosy-Westphal A, Bielfeldt F, Settler U, Monig H, Muller MJ. Measurement of waist circumference at four different sites in children, adolescents, and young adults: concordance and correlation with nutritional status as well as cardiometabolic risk factors. Obes Facts 2008;1: Burrows R, Gattas V, Leiva L, Maggi, Jiadrosic A, Lera L, Albala C. Perfil metabólico de riesgo cardiovascular en niños y adolescentes obesos con menor sensibilidad insulínica. Rev Med Chil 2005;133: Burrows RL, Burgueño M, Maggi A, Giodrosic V, Diaz E, Lera L, Albala C. Sensibilidad insulínica en niños de 6 a 15 años: asociación con estado nutricional y pubertad. Rev Med Chil 2006;134: Cruz ML, Weigensberg MJ, Huang TTK, Ball G, Shaibi GQ, Goran MI. The metabolic syndrome in overweight Hispanic youth and the role of insulin sensitivity. J Clin Endocrinol Metab 2004;89: Queiroz VM, Moreira PV, Vasconcelos TH, Toledo